Data Lab node
Histogram
The distribution of one channel’s values over the window: counts per bin (Bins 0 = Freedman–Diaconis auto), or a density; KDE adds a Gaussian kernel estimate (Silverman bandwidth) instead of bars. The Signal runs along the value axis in the input’s unit — Plot draws it as Bars or a Line — with Mode (the most common value) and Entropy in bits.
Inputs
| Port | Type | Default |
|---|---|---|
| SignalThe Data Lab signal to analyse — a Signal node's output, a Window, or any float array with its sample rate. Multi-channel signals are handled per channel. | structure | — |
| ChannelThe channel whose values are counted. | int | 0 |
| BinsNumber of bins; 0 = Freedman–Diaconis from the data’s spread. | int | 0 |
| Range AutoValue range from the data (it grows at once and shrinks slowly so the bars do not jump). | bool | true |
| Range MinManual value range, low end. | float | -1 |
| Range MaxManual value range, high end. | float | 1 |
| DensityProbability density (area 1) instead of counts. | bool | false |
| KDEA smooth Gaussian kernel estimate (Silverman’s bandwidth) instead of bars — plot it as a Line. | bool | false |
| Update RateHow many times per second the analysis is recomputed (default 10); 0 = every frame. Lower rates save CPU on long windows — a moved parameter always recomputes at once. | float | 20 |
Outputs
| Port | Type |
|---|---|
| SignalThe Data Lab signal to analyse — a Signal node's output, a Window, or any float array with its sample rate. Multi-channel signals are handled per channel. | structure |
| Counts | float |
| Centers | float |
| Mode | float |
| Entropy | float |
| Bin Width | float |
How It Works
Histogram shows the distribution of one channel's values over the window. Bins at 0 chooses the count automatically by the Freedman–Diaconis rule; Density normalises so the area is 1; KDE replaces the bars with a Gaussian kernel estimate at Silverman's bandwidth.
The important structural point: the output Signal runs along the value axis, in the input's own unit, not along time. Plot draws it as Bars or a Line with the value on the horizontal axis, which is what a histogram should look like.
Mode reports the most common value and Entropy the distribution's entropy in bits — a single number for "how spread out is this".
Use Cases
- Is this signal well behaved? — A glance at the distribution answers questions no mean and std can.
- Find the resting value — Mode is the value the signal actually sits at most of the time.
- Set a threshold honestly — Look at the distribution before picking an Events threshold.
- Measure variety — Entropy as a control: a flat distribution and a peaked one feel different.
- Compare two sources — Two histograms on one Plot shows a difference in character, not just in level.
Tips
- Leave Bins at 0 to start; the automatic rule is usually better than a round number you picked.
- KDE is the right choice for a smooth, publishable picture; bars are the honest one for small samples.
- Turn Range Auto off when you want two histograms on the same axis to be comparable.
- Density rather than counts when the window length can change, or the bars will grow with the window.
- A bimodal histogram is a signal telling you it has two states — often the most useful thing this node reports.
Open Histogram in the interactive catalog →
Histogram is one of 381 nodes in Grafit Pro.
Free download. Every feature for 7 days, no card. Then €99.99 once — or keep playing your projects as a Player.